Ultrasonic level controls

I am to the point where I would like to have more data about how full or empty a few tanks are. Currently, liquid level controls are used, just a probe that tells you if the liquid is at a certain point or not. Works fine most of the time, but leaves things to be desired.

Applications:

1. Milk tanks. Two seperate tanks. Targeted volume is 48-50,000 lbs. +/- 5500-5800 gallons, 22,000 liters. The tanks are horizontal, and do not have a round shape, rather somewhat egg shaped. Not a problem, as I have calibration charts for both tanks. The tanks are approximately 7 and a half feet, 228 cm tall. The application needs to be food grade, and while it won't be in contact with the milk, it will be subject to spraying of water and cleaning chemicals. I have a 3 inch stainless opening to work with, will require some custom fabrication, which I can do. Tank is indoors, but subject to frequent cleaning and water on the outside. Foaming is not too much of an issue. Accuaracy does not need to be dead on either. Pressure transducers aren't feasible, as they require welding into the bottom of a double walled insulated tank, and would be cost prohibitive. Temperature of the milk is also fairly constant.

2. Water tanks. Two seperate tanks, 20,000 gallons each, or 75,000 liters. Tanks are about 16 feet, or 5 meters tall. Tanks are vertical, same diameter all the way. Here accuracy is of much less concern, too bad it's an easier shape and liquid. Not food grade. Have a 4 inch FNPT threaded hole in the dead center of the top of the tank, which is outdoors and subject to the elements.

Will be interfacing these with Idec MicroSmart PLC's, I can get analog inputs for them.

What kind of sensors do I need to be looking at? Brands, suppliers, etc.

How do I integrate the funny shaped tank's calibration into the programming? Again, don't need to know exact volumes, especially when down low in the tank. As the tank reaches full capacity, accuracy needs to go up. The charts I have tell me the volume of milk by milimeter in the tank, so that data is available to me. The 2 tanks have slightly different callibration, and as such, well need different parameters.

Thanks for the inputs!

You may want to look at sensor such as a Ametek level probe. It gives a milliamp output. I am not sure if they make them for food grade though. Also do a search here on level probes, It's been awhile but there were some discussions on that.

Bolt,

Here are a couple of links to tank volume threads
http://www.plctalk.net/qanda/showthread.php?t=25255&highlight=tank+volume
http://www.plctalk.net/qanda/showthread.php?t=25694&highlight=tank+volume
http://www.plctalk.net/qanda/showthread.php?t=1397&highlight=horizontal
The last one is even for milk! You might also want to consider a pressure transmitter with a sanitary seal if you have a bottom port that is not subject to pump disturbance.

Well as stated earlier, I'm trying to get away from probes. Especially in the milk tank, it's not very handy. Pressure transducers are an accurate way, but our tanks don't have an extra fitting down low. I have seen places where they have them, and are quite pleased with their accuracy.

Those other threads are leading me in a few directions, so that should get my brain going.

Would it work to do a "table" of sorts in the PLC, so take the reading (eg. 15 ma) and "look" it up in a table to correspond with say 38,000 lbs, and then have 16 ma be 39,500 and if it was 15.5 ma, it would be 38,750? The tank is a very funny shape, and not sure I can figure out a formula on it. I will try though, maybe something will jump out at me. Also, tanks are on an incline, so they empty fully, so this will add to the funny characteristics of it. But I do have a very accurate and detailed chart, from 37 cm, which is 2748 lbs, and all the way to 228, which is 54967 lbs. The important info is in the top, we target 49,000 lbs, and it would be nice to know from 35,000 lbs up, until it gets there reasonably accurate. That way a full tank time can be estimated, etc.

Bolt,

There are well-known mathematical formulas that relate the volume of a cylindrical tank to the height of the liquid. Basically two formulas are used, one for liquid level 1/2 or below, and another for above 1/2. I once did a PLC program that computed the gallons of diesel fuel in a horizontal cylinderical tank, knowing the height of the fuel in the tank. The height was transmitted from a pressure tranducer, calibrated to convert pressure to liquid level.

The formulas for an egg-shaped tank would be different, but not impossible. It might take 4 formulas instead of two. Otherwise, you are looking at setting up a data look-up table. That is simple, but not usually as accurate as an exact forumula.

Here are the rung comments from my program. I used two rungs, one that computed the volume when the level was less than or equal (LEQ) to 2.75, and one when level was greater than 2.75. Note that the Arc Cosine (Cos-1, Inverse Cosine, or the ACS instruction) is used, a function available on the SLC 5/03 (and up. Only the last line in each paragraph below for "Gallons" is needed in the PLC calculations. The rest are merely explanations of how I got there:

If Diesel Tank Level is 1/2 full or less, compute Volume with following formula:
Inside Tank Radius = 2.75 feet
Length = 12'-5/8" = 12.0520833 feet
Area = R^2[Cos-1(R-h/R)
Volume = Area x Length
Gallons = 7.48052 gals per FT3 X Volume
Volume = Area x Length = 12.0520833 x 2.75 ^2 x 7.48052 x (Cos-1(R-h/R) = 12.0520833 X 7.5625 X 7.48052 x (Cos-1(R-h/R) =
Volume = Area x Length = 681.80359 x (Cos-1(R-h/R)
Measured Variable = FUEL PRESSURE as substitute for Fuel Height
Gallons = 681.80359 x {Inverse Cosine[2.75-H) / 2.75]}

If Diesel Tank Level is more than halfway, compute Volume with following formula:
Measured Variable = Fuel Height
Inside Tank Radius = 2.75 feet
Length = 12'-5/8" = 12.0520833 feet
Area = R^2 [PI - Cos-1(h-R/R)]
Volume (FT3) = Length X Area
Gallons = 7.48052 gals per FT3 X Volume (FT3) = 7.48052 X (Length X Area)
Area = (2.75 ^2 x 3.14159) - [2.75^2 x (Cos-1((h-R)/R)]
Area = 23.7583 - 7.5625{Inverse Cosine[(H-2.75) / 2.75]}
Gallons = 7.48052 X 12.0520833 X Area = 90.155847 X Area = 90.155847 X (23.7583 - 7.5625{Inverse Cosine[(H-2.75) / 2.75]})

Bolt said:

Would it work to do a "table" of sorts in the PLC, so take the reading (eg. 15 ma) and "look" it up in a table to correspond with say 38,000 lbs, and then have 16 ma be 39,500 and if it was 15.5 ma, it would be 38,750?

Click to expand...

Yes, check the threads I listed

Bolt said:

Also, tanks are on an incline, so they empty fully, so this will add to the funny characteristics of it.

Click to expand...

It's discussed in the threads

Bolt said:

The important info is in the top, we target 49,000 lbs, and it would be nice to know from 35,000 lbs up, until it gets there reasonably accurate. That way a full tank time can be estimated, etc.

Click to expand...

Tom Jenkins has the example you need in the last thread I listed.

I reply to this topic, check out the other post, and realize most of my questions have been answered. Doh! Anyways, how about the the hardware part, what brands and models do I need to be looking at?

Someday, when I am bored, I should enter the calibration charts into Excel, and see what kinda curve/formula I can come up with. Probably a different formula for different sections of the tank. To make it even funner, the 2 tanks aren't the same in calibration.

The charts give me a lbs and liters for every milimeter of milk in the tank, from 307 mm to 2280 mm. Hmmm, close to 2000 sets of numbers. Breaking it down into managable chunks will help me here. Anything over 193 centimeters is overfull, so it doesn't matter from that point up. Alarm will be going off regardless of level. Anything below 140 cm isn't of much use for data yet. So that makes it a smaller window where accuracy is needed. I can run different formulas for different zones of the tank, and go from there.

Miltronics made an excellent ultraisonic transduer, but I believe that Siemens bought them out. I am not sure if the old technology was brought over with the purchase (something to look into).

Be very careful with the installation of ultrasonics as they do not work well in a misty application. They also do not like moisture on their face.

I would consider leaving the current level probe in for High High level alarm (seen to many overflows).

Hi,
We use some VEGA Ultrasonic/Radar sensors on site, the set up software has all different shapes of tanks in it, you just input your diamentions, or I think there is an option to just fill out a table like the info you have. You then get an mA output proportional to the volume of liquid in the tank.
These units are quite expensive, but for non contact level measurement they are very good.

http://www.vega.com/en/index_us.htm

My worst application was a Hydrofloric acid tank. We solved that with an Endress +Hauser radar unit. Check with a local E +H rep for sanitary use. For the water you need to be careful about sensor placement. Some units don't like dead-center postion due to the additional reflected echoes strength.

Can you put on some load cells for the milk tank? Weight is a good way for an odd sized angled tank. Kistler-Morse http://www.kistlermorse.com/p_lcell.asp makes units for retrofit I believe. If you use weight you wouldn't need to worry about some formula to fit an odd shape or cleaning issues.

Just a thought since the one tank is odd shaped and rests at an incline WHY not use flow meters? If you know how much you put in and how much you take out, then you should always know how exactly how much is in the tank. This would remove the need for formulas, it would be simple math.

To add even more accuracy to Ron's flowmeter concept keep the present probe point and use the transition point of this contact to re-reference your in and out flow balance. This would cancell out any flowmeter errors. As the probe makes (or breaks) this would be a known volume. You would use this known volume to recalibrate your calculated volume from the flowmeters.

milldrone said:

To add even more accuracy to Ron's flowmeter concept keep the present probe point and use the transition point of this contact to re-reference your in and out flow balance. This would cancell out any flowmeter errors. As the probe makes (or breaks) this would be a known volume. You would use this known volume to recalibrate your calculated volume from the flowmeters.

Click to expand...

Excellent idea, then you don't need to program the formula for volume of a slanted cylindrical tank into your PLC...just take a one time set of calculations after installing (or verifying existing) flowmeter calibration.

If there are not already flowmeters, I would recommend you try an analog teachable ultrasonic....(Banner QT50)

Have the operator clean it daily, and verify visually it's performance...

I have used some Banner ultrasonics that were super reliable and easy to set up...my product being sensed was radically differnet than yours however it was equally difficult...

Much less money than flowmeters...

EDIT: This one is excellent: http://info.bannersalesforce.com/xpedio/groups/public/documents/datasheet/70137.pdf